The Cytokinesis Formins from the Nematode Worm and Fission Yeast Differentially Mediate Actin Filament Assembly*

Formins drive actin filament assembly for diverse cellular processes including motility, establishing polarity, and cell division. To investigate the mechanism of contractile ring assembly in animal cells, we directly compared the actin assembly properties of formins required for cytokinesis in the nematode worm early embryo (CYK-1) and fission yeast (Cdc12p). Like Cdc12p and most other formins, CYK-1 nucleates actin filament assembly and remains processively associated with the elongating barbed end while facilitating the addition of profilin-actin above the theoretical diffusion-limited rate. However, specific properties differ significantly between Cdc12p and CYK-1. Cdc12p efficiently nucleates filaments that in the presence of profilin elongate at approximately the same rate as control filaments without formin (∼10.0 subunits/s). CYK-1 is an inefficient nucleator but allows filaments to elongate profilin-actin 6-fold faster than Cdc12p (∼60 subunits/s). Both Cdc12p and CYK-1 bind to pre-assembled actin filaments with low nanomolar affinity, but CYK-1 dissociates 2 orders of magnitude more quickly. However, CYK-1 rapidly re-associates with free barbed ends. Cdc12p allows barbed ends to elongate in the presence of excess capping protein, whereas capping protein inhibits CYK-1-mediated actin assembly. Therefore, these evolutionarily diverse formins can drive contractile ring assembly by a generally similar mechanism, but cells with unique dimensions and physical parameters might require proteins with carefully tuned actin assembly properties.

The final step of cell division is cytokinesis, the physical separation of a mother cell into two daughter cells (1,2). Animal cells spatially and temporally coordinate cleavage site positioning through both astral and spindle microtubules (3,4). Upon choosing a division site, far less is known about how actin and myosin II motor filaments assemble into the contractile ring. The mechanism is clearer in the unicellular fission yeast Schizosaccharomyces pombe (see Fig. 7A) (5). The fission yeast contractile ring is constructed from ϳ60 "pre-ring" nodes through the coordinated activities of numerous proteins including the anillin-like protein Mid1p (6 -8), the type II myosin motor Myo2p (9 -11), the actin monomer binding protein profilin Cdc3p/SpPRF (12), and the actin filament nucleator formin Cdc12p (13).
A major question is whether animal cells assemble the contractile ring by a similar mechanism as fission yeast. Many of the proteins required for cytokinesis are conserved between fission yeast and animal cells including anillin, type II myosin, profilin, and formin (1,2). Formins are large multidomain proteins that, in addition to cytokinesis, assemble actin filaments for multiple cellular processes including motility and establishing polarity (14 -16). Formins contain a highly conserved actin assembly FH2 2 domain, and associated proline-rich profilin binding FH1 domain, which are flanked on either side by regulatory domains.
The formin family is large and diverse with at least 18 formin genes in mammals, six each in the fruit fly Drosophila melanogaster and the nematode worm Caenorhabditis elegans, and three in fission yeast S. pombe (17). A prevalent model is that multiple formin isoforms are required because each assembles actin for a distinct cellular process. The "one formin per cellular role" paradigm is certainly true in fission yeast where each of the three isoforms has a distinct role: Cdc12p, cytokinesis; For3p, polarized growth; Fus1p, mating (18 -20). The specific cellular role(s) of most formins are not known, and the molecular basis for functional specificity is unclear. Activating a particular formin isoform at the right time and place via the terminal regulatory domains is certainly important for functional specialization (15,21). Additionally, the rates of actin assembly may also be important. Several properties differ significantly between diverse formins, such as the actin filament elongation rate, which can vary almost 10-fold (22)(23)(24)(25)(26).
The leading model for animal cell cytokinesis is the nematode worm C. elegans early embryo (27). C. elegans may assemble the contractile ring by a similar mechanism as fission yeast because the formin CYK-1 and profilin CePFN-1 are absolutely required for cytokinesis (28 -30). However, the biochemical activities of most worm embryo cytokinesis proteins are not known.
We directly compared the actin assembly properties of the nematode worm cytokinesis formin CYK-1 and the fission yeast cytokinesis formin Cdc12p to gain molecular insight into the mechanism of contractile ring assembly in animal cells and to test the hypothesis that the specific actin assembly rates are conserved between evolutionarily diverse formins with similar cellular roles. An active CYK-1 construct containing the actin assembly domains CYK-1(FH1FH2) nucleates actin filament assembly, remains processively associated with the elongating actin filament barbed end, and drives the addition of actin bound to profilin CePFN-1 above the theoretical diffusion-limited rate. However, the specific rates of CYK-1 are significantly different from Cdc12p. Cdc12p efficiently nucleates actin filaments that elongate slowly. CYK-1 is an inefficient nucleator but has a high affinity (low nanomolar) for preassembled actin filament barbed ends and drives exceptionally fast actin monomer addition. Therefore, the general mechanism of formin-mediated contractile ring assembly may be similar between fission yeast and the nematode worm embryo. However, differences in their specific actin assembly properties may tune these formins for contractile ring assembly in diverse cells.
Ca-ATP actin was purified from chicken skeletal muscle (Trader Joe's) as described for rabbit skeletal muscle (34). Gelfiltered actin was labeled on Cys-374 with pyrenyl iodoacetamide or Oregon green 488 iodoacetamide (Invitrogen) (23,35). Tetramethylrhodamine-labeled actin was a gift from Ron Rock (The University of Chicago). Immediately before each experiment, 5-15 M Ca-ATP actin was converted to Mg-ATP actin by adding 0.1 volume of 2 mM EGTA and 0. 5  Fluorescence Spectroscopy-Actin assembly was measured from the fluorescence of a trace of pyrene-actin (excitation at 364 nm and emission at 407 nm) with Spectramax Gemini XPS (Molecular Devices) and Safire 2 (Tecan) fluorescent plate readers. Final protein concentrations are indicated in the figure legends.
For spontaneous assembly assays, a 15 M mixture of pyrenelabeled and unlabeled Mg-ATP-actin with 100X anti-foam 204 (0.005%; Sigma) was placed in a upper row of a 96-well nonbinding black plate (Corning). Other proteins to be assayed (formin, profilin, etc), 10ϫ KMEI (500 mM KCl, 10 mM mgCl 2 , 10 mM EGTA, and 100 mM imidazole, pH 7.0) and Mg-buffer G (2 mM Tris, pH 8.0, 0.2 mM ATP, 0.1 mM MgCl 2 , and 0.5 mM DTT) were placed in a lower row of the plate. Reactions were started by mixing contents of the lower wells and the actin monomers in the upper wells with a 12-channel Pipetman (Eppendorf).
For seeded assembly assays, 5.0 M unlabeled Mg-ATP-actin was preassembled in the upper row of the plate followed by addition of anti-foam, other proteins to be assayed (formin, profilin, etc.), and Mg-buffer G. A 5.0 M mixture of pyrenelabeled and unlabeled Mg-ATP-actin with Mg-buffer G was placed in the lower plate row. Mixing actin monomers in lower wells with preassembled actin filaments in upper wells started reactions.
For depolymerization assays, a 5.0 M mixture of unlabeled and pyrene-labeled Mg-ATP-actin monomers was preassembled in the upper row of the plate for 2 h followed by the addition of anti-foam. Formin, 10ϫ KMEI, and Mg-buffer G were placed in the lower plate row. Reactions were started by mixing lower wells with upper wells, diluting the pre-assembled filaments to 0.1 M.
The critical concentration for actin assembly was determined by assembling a 1.0 M mixture of unlabeled and pyrene-labeled Mg-ATP-actin monomers in the presence of a range of concentrations of formin in a 96-well black plate. The final amount of filamentous actin was determined after a 16-h incubation in the dark at 25°C.

Calculation of Initial Polymerization Rates, Depolymerization Rates, Barbed End Affinity, and Nucleation Efficiency-
The actin assembly rates from spontaneous assembly reactions were determined by measuring the slopes from the points where 10 -50% of the actin had assembled. Polymerization rates from pre-assembled actin filament seeds were measured from the slope of a linear fit of the first 300 s. The rate of depolymerization was calculated by fitting the data from 300 to 1000 s with a single exponential curve. Depolymerization rates were expressed as a percent normalized to the rate of actin alone. The affinity of formin for actin filament barbed ends was determined by fitting plots of the dependence of either the initial assembly rate or the initial depolymerization rate on the concentration of formin, with the equation [formin])/2[ends])) 1 ⁄2, where Vi is the observed elongation or depolymerization rate, Vif is the elongation or depolymerization rate when barbed ends are free, Vib is the elongation or depolymerization rate when barbed ends are bound, [ends] is barbed-end concentration, and [formin] is formin concentration (32). The nucleation efficiency was calculated by dividing the spontaneous assembly rate (slope) by kϩ [where kϩ ϭ 0.3 and 10.6 M Ϫ1 s Ϫ1 for Cdc12(FH1FH2)p in the absence and presence of profilin and 6.2 and 63.2 M Ϫ1 s Ϫ1 for CYK-1(FH1FH2) in the absence and presence of profilin] and then dividing by the formin concentration.
Microscopy of Fluorescently Labeled Filaments-Products of spontaneous assembly reactions were examined by fluorescence microscopy as described previously (23,38). Actin filament annealing (39) was examined by assembling a blend of unlabeled Mg-ATP-actin with either rhodamine-labeled Mg-ATP-actin or Oregon green-labeled Mg-ATP-actin. Mixtures of equal amounts of red and green filaments were sheared by pushing 16 times through a 3/8-inch 26-gauge needle on a 1.0-ml tuberculin syringe and allowed to anneal for 60 min at room temperature. Reactions were terminated by a 250-fold dilution in fluorescence buffer (50 mM KCl, 1 mM MgCl 2 , 100 mM DTT, 20 g/ml catalase, 100 g/ml glucose oxidase, 3 mg/ml glucose, 0.5% methylcellulose, 10 mM imidazole, pH 7.0) and absorbed to coverslips coated with 0.05 g/l poly-L-lysine. Fluorescence images were collected with a cooled CCD camera (Orca-ER) on an Olympus IX-81 microscope.

The Worm Cytokinesis Formin CYK-1 Reduces the Rate of Monomer Association and Dissociation from the Actin Filament
Barbed End-Most formins bind processively to actin filament barbed ends with low nanomolar affinity and in the absence of profilin lower the rate of both assembly and disassembly (16). We began our investigation of the worm cytokinesis formin CYK-1 by determining whether CYK-1 binds actin filamentbarbed ends and influences monomer association and dissociation and whether the specific rates are similar in the presence of CYK-1 and the fission yeast cytokinesis formin Cdc12p.
A fission yeast Cdc12p construct containing both the formin homology 1 and 2 domains Cdc12(FH1FH2)p as well as a construct lacking the profilin binding FH1 domain Cdc12(FH2)p shifts the critical concentration for assembly from that of the barbed end near 0.1 M to that of the pointed end near 0.9 M (Fig. 1A) (23,32). A range of concentrations of CYK-1(FH1FH2) and CYK-1(FH2) constructs had no effect on the critical concentration (Fig. 1A), suggesting that CYK-1 does not strongly inhibit barbed end assembly.
CYK-1 Inefficiently Nucleates Actin Filament Assembly-Although formins reduce the rate of barbed end addition (Figs. 1, B and C) (16), formins enhance the overall rate that actin monomers assemble into filaments by stimulating nucleation. CYK-1(FH1FH2) increases the spontaneous assembly of Mg-actin monomers by reducing the lag at the outset of the reaction and increasing the maximum rate (slope) of assembly ( Fig. 2A) (23). The general ability of CYK-1(FH1FH2) and Cdc12(FH1FH2)p to stimulate spontaneous actin monomer assembly is similar over a range of formin concentrations (Fig. 2B). Constructs containing only the formin homology 2 domain CYK-1(FH2) and Cdc12(FH2)p also stimulate the spontaneous assembly of actin monomers (Fig. 2B). Because CYK-1 does not shift the critical concentration for assembly near that of the pointed end (Fig. 1A), CYK-1(FH1FH2) enhances spontaneous actin assembly at lower actin concentrations than Cdc12(FH1FH2)p (Fig. 1E).
The rate of spontaneous actin assembly is dependent upon both the number of filaments (nucleation) as well as the rate that the filaments assemble (elongation). CYK-1 allows actin filaments to elongate significantly faster than Cdc12p (Fig. 1C), suggesting that Cdc12(FH1FH2)p is a more efficient nucleator than CYK-1(FH1FH2) since the spontaneous actin assembly activity of these formins is similar (Fig. 2B). To compare the relative nucleation efficiency of CYK-1 and Cdc12p, we looked at the length of filaments after the spontaneous assembly reactions shown in Fig. 2A reached a plateau. Fluorescence images of the products of the spontaneous actin assembly reactions labeled with rhodamine-phalloidin showed that filaments formed in the presence of CYK-1 constructs are ϳ9ϫ longer than filaments formed in the presence of Cdc12p constructs, indicating that CYK-1 is a less efficient nucleator ( Knowing that CYK-1-nucleated filaments elongate their barbed ends at ϳ6.0 subunits s Ϫ1 M Ϫ1 (see Fig. 5; Table 1) and Cdc12p nucleated filaments elongate their barbed ends at ϳ0.2 subunits s Ϫ1 M Ϫ1 (see Fig. 5; Table 1) (22)(23)(24), we calculated the nucleation efficiency of CYK-1 and Cdc12p (Fig. 2D). At low concentrations CYK-1(FH1FH2) and CYK-1(FH2) maximally produced 1 new filament per ϳ50 and ϳ80 molecules (Fig. 2D). This is very inefficient compared with Cdc12(FH1FH2)p and Cdc12(FH2)p, which maximally produced 1 new filament per ϳ2.5 and ϳ3.0 molecules at low concentrations ( Fig. 2D) (23).
The effect on spontaneous actin assembly of a range of concentrations of either CePFN-1 with CYK-1(FH1FH2) or fission yeast profilin SpPRF with Cdc12(FH1FH2)p is biphasic (Figs. 3, B and C) (23). Low profilin concentrations increasingly enhance the spontaneous assembly rate until an optimal con-  centration of ϳ2.5 M is reached. Higher profilin concentrations decrease the assembly rate because profilin inhibits nucleation, and profilin unassociated with actin effectively excludes profilin-actin from formin (22,40).
The Rate of CYK-1-mediated Seeded Assembly Is Increased by Profilin-Although profilin reduces the nucleation efficiency of formin (Fig. 3D), profilin increases the "bulk" spontaneous actin assembly rate in the presence of both Cdc12(FH1FH2)p and CYK-1(FH1FH2) (Figs. 3, A-C). For Cdc12(FH1FH2)p and other formins, this paradox is because profilin significantly increases the barbed end elongation rate of formin nucleated actin filaments (16). We, therefore, investigated how a range of profilin concentrations affects the addition of Mg-ATP actin monomers to the barbed ends of preassembled actin filaments in the presence of formin(FH1FH2) and formin(FH2) constructs (Figs. 4, A and B).
Capping Protein Inhibits Actin Assembly in the Presence of CYK-1-Most formins inhibit capping protein, allowing the assembly of long actin filaments (16). We initially compared the ability of capping protein to inhibit the addition of profilin-Mg-ATP actin monomers to pre-assembled actin filaments in the     Cdc12(FH1FH2)p with SpPRF and CYK-1(FH1FH2) with CePFN-1 (Figs. 4, E and F). With Cdc12(FH1FH2)p and SpPRF, capping protein concentrations up to 500 nM had little effect on both the spontaneous assembly rate as well as the average filament length 5 min into the reaction. However, with CYK-1(FH1FH2) and CePFN-1, capping protein inhibited the spontaneous assembly rate and reduced the average filament length near that of Cdc12(FH1FH2)p and SpPRF.

Real-time Visualization of Actin Assembly in the Presence of CYK-1 by Total Internal Reflection Fluorescence Microscopy-
We utilized time-lapse evanescent wave fluorescent microscopy to directly observe formin-mediated actin assembly in real-time to determine 1) whether CYK-1(FH2) and CYK-1(FH1FH2) remain processively associated with the elongating barbed end in both the absence and presence of profilin, 2) the specific elongation rates of CYK-1(FH2)-and CYK-1(FH1FH2)-associated actin filament barbed ends with and without profilin, and 3) the rate that CYK-1(FH1FH2) dissociates from the actin filament barbed end (Figs. 5 and 6; Table 1). We followed the assembly of individual filaments elongating from a pool of 1.0 M Mg-ATP-actin monomers supplemented with a trace of 0.5 M Mg-ATP-actin monomers labeled with Oregon green for visualization (22,24,35).
In the absence of formin all filaments elongate their barbed ends at the same constant rate of ϳ9.0 subunits/s (Figs. 5, A and D; Table 1) (22,24). As seen previously with Cdc12p (22)(23)(24), two distinct filament populations are detected in both the absence and presence of profilin (Figs. 5, E and F). The first population consists of internal control filaments that elongate at ϳ9.0 subunits/s, the same rate as filaments in the absence of formin. The second population consists of Cdc12p-associated filaments that elongate at a significantly different rate, which depends upon profilin SpPRF (Figs. 5, E and F). In the absence of SpPRF, Cdc12(FH1FH2)p-nucleated filaments elongate at only 0.3 subunits/s (Fig. 5E). Four micromolar SpPRF increases the elongation rate of Cdc12(FH1FH2)p-nucleated filaments to 10.6 subunits/s (Fig. 5F). Filaments nucleated by Cdc12(FH2)p, which lacks the profilin binding FH1 domain, elongate at 0.26 and 0.02 subunits/s in the absence and presence of 4.0 M SpPRF (Fig. 5, G and H). The presence of two filament populations verifies processive association of Cdc12p with the elongating barbed end. If formin rapidly came on and off, only one filament population would be detected that elongates at an intermediate rate (22,24).
Two filament populations are also present with the CYK-1 constructs, confirming processive association with the elongating barbed end in both the absence (Figs. 5, B and I) CePFN-1 (Figs. 5, K and L).
The barbed-end elongation rate in the presence of both CYK-1(FH1FH2) and Cdc12(FH1FH2)p has a biphasic dependence on the concentration of profilin (Fig. 6G) (22). Lower profilin concentrations increase the elongation rate, with maximal effect in the range of 2.5-5.0 M profilin. However, elongation is increasingly inhibited at higher profilin concentrations. CYK-1(FH1FH2)-associated actin filament barbed ends elongate significantly faster than Cdc12(FH1FH2)p-associated barbed ends at all profilin concentrations.
CYK-1 Dissociates More Rapidly Than Cdc12p from the Elongating Barbed End-In the absence of formin, all filaments visualized by total internal reflection fluorescence microscopy that assemble from a pool of 33% Oregon greenlabeled Mg-ATP actin monomers elongate at a constant rate and are uniformly bright (Fig. 5A). In the presence of formin, two filament populations are detected that differ by elongation rate because formin remains processively associated with the elongating barbed end and influences actin monomer addition (Figs. 5, B and C). In the presence of formin and profilin, rapidly growing formin-associated filaments are also less bright than control filaments because profilin selects against the labeled actin (Figs. 6, A-E) (5,22). Switches from a dim fast filament to a bright slow filament provide a convenient method to identify CYK-1 dissociation events (Fig. 6A). Kymographs of the length of individual filaments over time allow easy visualization of switches between periods of formin-dependent fast dim elongation and formin-independent bright slow elongation, apparent formin dissociation events (Figs. 6, B-D). A plot of the fraction of filaments bound to formin versus time shows that on average both Cdc12(FH1FH2)p and CYK-1(FH1FH2) allow the addition of thousands of subunits before dissociating (Fig. 6F). However, CYK-1(FH1FH2) dissociates 2 orders of magnitude more rapidly than Cdc12p: 7.1 ϫ 10 Ϫ5 s Ϫ1 for Cdc12(FH1FH2)p and 3.9 ϫ 10 Ϫ3 s Ϫ1 for CYK-1(FH1FH2).
Although CYK-1 dissociates more rapidly than Cdc12p, both formins have a low nanomolar affinity for the actin filament barbed end (Figs. 1, C and E), suggesting that the barbed end association rate constant for CYK-1 must be high. In agreement, CYK-1(FH1FH2) molecules rapidly re-associate with free actin filament barbed ends revealing filaments with multiple alternating stretches of dim fast-growing segments and bright slow-growing segments (Figs. 6, D and E).
As another measure of the dissociation rate, we also compared the ability of formins to inhibit end-to-end actin filament annealing (Figs. 6, H and I) (23,39). Mixtures of sheared red (rhodamine-labeled) and green (Oregon green-labeled) filaments will anneal over time resulting in longer filaments that alternate between red and green segments (Fig. 6H) (23). Five min after shearing, actin filaments average 1.0 m in length. After 60 min of annealing, control actin filaments without formin average 3.8 m in length. Annealing is strongly inhibited by 250 nM Cdc12(FH1FH2)p in both the absence (0.96 m) and presence of 5.0 M profilin SpPRF (0.95 m) (Fig. 6H). However, annealing is only partially inhibited by 250 nM CYK-1(FH1FH2) in both the absence (2.4 m) and presence of 5.0 M profilin CePFN-1 (2.3 m) (Fig. 6H). A plot of the dependence of the actin filament length on the concentration of formin shows that high concentrations of CYK-1(FH1FH2) allow annealing to occur (Fig. 6I). CYK-1(FH1FH2) may allow annealing by switching between bound and unbound states.

DISCUSSION
By direct comparison, we discovered that the actin assembly properties are generally similar between the cytokinesis formins from the nematode worm C. elegans CYK-1 and fission yeast S. pombe Cdc12p. However, various rate constants differ significantly. Therefore, the general mechanism of contractile ring assembly in fission yeast and the nematode worm early embryo may be similar, but cells with different dimensions and physical characteristics require proteins with carefully tuned actin assembly properties.

Actin Assembly Properties of the Cytokinesis Formins Cdc12p and CYK-1
Processive Barbed End Association-The key intrinsic property of formin is the ability to remain continually associated with the elongating actin filament barbed end. With the one exception (41), all formins including CYK-1 remain processively associated with the elongating barbed end through their FH2 domain in both the absence and presence of profilin (Refs. 15 and 16; Fig. 5). Processive association allows formins to variously modify the elongation rate. Without profilin fission yeast Cdc12p, mouse mDia2, budding yeast Bni1p, worm CYK-1, and mouse mDia1 reduce the barbed end elongation rate from 10.0 subunits/s to 0.2, 1.5, 5.0, 6.0, and 9.5 subunits/s (Table 1) (22).
To test whether differences in the specific elongation rate are important, the physiological consequences of replacing the FH1FH2 domains from a "slow" formin like Cdc12p with the FH1FH2 domains from CYK-1 or another "fast" formin need to be determined. The specific actin assembly properties of formins with relevant actin isoforms also need to be determined. Considering that fission yeast profilin binds ϳ10-fold better to fission yeast actin than to muscle actin (43), Cdc12p might be able to increase the assembly rate of fission yeast actin significantly more than muscle actin in the presence of profilin. Nucleation Efficiency-Given that formin FH2 domains form homodimers (44), essentially each Cdc12p homodimer initiates assembly of a new filament. CYK-1 is a significantly less efficient nucleation factor, requiring ϳ25 CYK-1 homodimers for every new actin filament (Fig. 2D). Profilin reduces the nucleation efficiency of all formins. Cdc12p and CYK-1 require ϳ25 and ϳ225 homodimers to produce a new actin filament in the presence of profilin-actin (Fig. 3D).
Association with Preassembled Actin Filaments-Both Cdc12p and CYK-1 bind to the barbed end of preassembled filaments with low nanomolar affinity (Figs. 1, C and D). Formin isoforms like CYK-1 might circumvent inefficient nucleation by binding to filaments assembled by other actin nucleation factors such as the Arp2/3 complex or Spire. The "convergent elongation model," originally proposed to explain how vasodilator-stimulated phosphoprotein (VASP) uses Arp2/3 complex-nucleated filaments at the leading edge of motile cells to initiate the assembly of filopodia (45), might also explain the formin-dependent assembly of particular actin-based cellular structures.
Although Cdc12p-and CYK-1-associated barbed ends elongate at significantly different rates, both depolymerize ϳ80% slower than control filaments without formin (Figs. 1, C and E). Dissimilar elongation rates may reflect differences in the equilibrium between "closed" and "open" states as formins "walk" the elongating barbed end (22,40,46). Therefore, the reverse closed and open state transitions of formins during monomer dissociation may occur at similar rates. It is important to determine whether formins remain processively associated with a depolymerizing barbed end.
Barbed End Dissociation Rate-In vitro formins are able to facilitate the addition of thousands of actin subunits to the elongating barbed end. These filaments can be 100s of microns in length, significantly longer than typical actin filaments in cells. Therefore, formins must ultimately be turned off in cells. Formin-dependent actin assembly might be stopped by re-activated autoinhibition (15,21), the action of other factors such as capping protein, and/or the intrinsic dissociation rate of formin. CYK-1 dissociates from the barbed end 2 orders of magnitude more quickly than Cdc12p (Fig. 6F). The intrinsic dissociation rate may be enough to turn off CYK-1, whereas other mechanisms stall Cdc12p-mediated elongation.
Inhibition of Capping Protein-The ability of formin to allow barbed end seeded elongation in the presence of capping protein provided the first indirect evidence that formins remain processively associated with the elongating barbed end (32,(47)(48)(49). However, capping protein can inhibit barbed ends associated with the mouse formin mDia1 (26), suggesting that capping protein may have different effects depending upon the specific formin.
CYK-1 and Cdc12p both reduce the affinity of capping protein for actin filament barbed ends ϳ50-fold. However, capping protein completely blocks the stimulation of spontaneous actin monomer assembly and barbed end elongation in the presence of CYK-1 but not in the presence of Cdc12p (Figs. 4, D and F). The basis for this difference is not known, but it might simply reflect the 2 orders of magnitude faster barbed-end dissociation rate of CYK-1 compared with Cdc12p. Further work is required to determine the mechanism(s) by which Cdc12p is inhibited.
Actin Filament Bundling and Severing-In addition to nucleation and modification of the barbed end elongation rate, several formin isoforms are capable of binding to and bundling or severing actin filaments (41, 47, 50 -52). A physiological relevance for these activities has not been demonstrated, although formin-dependent cellular structures such as the contractile ring and filopodia are typically composed of bundled actin filaments. CYK-1 and Cdc12p do not bind and bundle actin filaments at any concentration, and neither CYK-1 nor Cdc12p effectively sever actin filaments at the nanomolar concentrations that are sufficient to nucleate actin filament assembly and bind preassembled actin filament barbed ends (data not shown).

Contractile Ring Assembly in Fission Yeast and the Nematode Worm Embryo
The fission yeast contractile ring assembles from ϳ60 prering nodes composed of at least seven proteins including the actin filament motor protein myosin II Myo2p (9 -11) and the formin Cdc12p (13). The coordinated effort of formin Cdc12pmediated actin filament assembly coupled with type II myosin Myo2p-mediated actin filament pulling (53) connects pre-ring nodes and drives their coalescence into a mature contractile ring (Fig. 7A) (5,54,55). Pre-ring node-associated Cdc12p nucleates the assembly of a filament whose pointed end is pushed away by insertional barbed end elongation. Myo2p on an adjacent node captures the actin filament pointed end, and Myo2p "walking" toward the Cdc12p-associated barbed end pulls the pre-ring nodes together (Fig. 7A) (5). Before contractile ring assembly Cdc12p has also been visualized in a larger progenitor "spot" that might also contribute to contractile ring assembly (13,56,57).
The mechanism of contractile ring assembly in animal cells might be similar to fission yeast. In the early nematode worm embryo, the actin motor myosin II NMY-2 localizes to discrete "foci" that coalesce into the contractile ring (Fig. 7B) (58,59). These foci may be analogous to the fission yeast pre-ring nodes as they are connected by filamentous actin (59). Although the formin CYK-1 localizes to the cleavage furrow (30) and CYK-1 and profilin CePFN-1 are required for early steps in contractile ring assembly (29), it is not known whether CYK-1 or other contractile ring proteins localize to the myosin II foci.
The general actin assembly properties of CYK-1 and Cdc12p are similar (Figs. 1-6), suggesting that the mechanism of contractile ring assembly in the nematode worm embryo may be similar to fission yeast. Differences in specific rates between evolutionarily diverse formins such as nucleation efficiency and elongation rate might be relevant for specializing actin assembly for cells with different physical parameters.
First, the fission yeast cortical actin cytoskeleton is sparse by comparison to animal cells, necessitating the need for an efficient nucleation factor such as Cdc12p to initiate the assembly of "new" filaments from pre-ring nodes. The Arp2/3 complex is not necessary for contractile ring assembly in fission yeast (55,60,61). Conversely, the worm embryo is filled with preassembled cortical actin filaments (59). CYK-1 might associate with preassembled cortical filaments and drive contractile ring assembly by a similar mechanism as Cdc12p. The source of preassembled filaments in the worm embryo is less clear. The actin related protein Arp2/3 complex is dispensable for actin filament assembly in the early embryo (29).
Second, the circumference of the worm embryo is ϳ6 times longer than fission yeast, which is similar to the difference in the barbed end elongation rate of profilin-actin in the presence of Cdc12p and CYK-1 ( Fig. 5; Table 1). The efficient assembly of larger contractile rings might require a faster rate of actin elongation. However, cell size and the rate of formin-mediated elongation do not always correlate. Budding yeast and fission yeast cells are similar sizes, but the budding yeast formin Bni1 mediates the assembly of filaments that elongate twice as fast as Cdc12p (22,24). It will be interesting to assess the consequence on the rate and efficiency of contractile ring assembly by modifying the formin-dependent actin filament elongation rate in an individual cell type.
Other factors, such as the number and length of individual filaments as well as stiffness of the cortex are also relevant parameters. Interestingly, because capping protein specifically inhibits CYK-1, in the presence of high capping protein concentrations the average length of filaments assembled with CYK-1 and Cdc12p are similar (Fig. 4D).
Third, a key parameter to contractile ring assembly in fission yeast is that pre-ring nodes are pulled together by transient connections rather than long-lived connections that favor clumping over coalescence (5). Given the slow intrinsic dissociation rate of Cdc12p, other factors are likely required to terminate connections between nodes, such as the actinsevering protein cofilin. The faster dissociation rate of CYK-1 might be sufficient to provide temporary connections between myosin foci during contractile ring assembly in the worm embryo (Fig. 6F).
To understand how myosin II foci coalesce, we need to determine the spatial and temporal localization of CYK-1 and other contractile ring components as well as precisely image actin filaments in real-time during contractile ring assembly in the nematode worm embryo. Live cell imaging coupled with additional biochemical characterization of the participating set of proteins will be crucial for developing testable mechanistic models of contractile ring assembly in animal cells.

A Fission Yeast B Worm Embryo
Actin Filaments ? FIGURE 7. Contractile ring assembly in fission yeast and the nematode worm embryo. A, the fission yeast contractile ring is assembled from ϳ60 pre-ring nodes which are composed of at least seven proteins including the formin Cdc12p and the molecular motor myosin II (54,55). Formin Cdc12p-mediated actin filament assembly coupled with myosin II-mediated actin filament pulling links pre-ring nodes and ultimately drives their coalescence into a mature contractile ring (5). B, similar to fission yeast pre-ring nodes, the actin molecular motor myosin II NMY-2 localizes to discrete foci that coalesce into the contractile ring in the nematode worm embryo (58,59). The worm embryo is full of preassembled actin filaments that could be utilized by the formin CYK-1 to drive contractile ring assembly. However, the localization of CYK-1 is unclear, and other myosin II foci components are not known.